Tensorflow model.evaluate() 崩溃，因为不支持 label_mode 中的 None 值

Question

我尝试在 preprocessing.image_dataset_from_directory 上运行 model.evaluate() 但无济于事，因为 label_mode=None

我正在尝试从 ImageDataGenerator 的 flow_from_directory 实现与 class_mode='input' 类似的功能。我已尝试多次并不断收到相同的错误消息。我也尝试过手动更改模型的输入，但我仍然不确定哪里出了问题。下面是我的代码：

    SIZE = 128
    batch_size = 64

    train_generator = preprocessing.image_dataset_from_directory(
        r'C:\Users\#omitted user name#\Downloads\archive (1)\noncloud_train',    
        image_size=(SIZE, SIZE),
        batch_size=batch_size,
        label_mode=None
   )

   validation_generator = preprocessing.image_dataset_from_directory(
       r'C:\Users\#omitted user name#\Downloads\archive (1)\noncloud_test',
       image_size=(SIZE, SIZE),
       batch_size=batch_size,
       label_mode=None

   )

  anomaly_generator = preprocessing.image_dataset_from_directory(
      r'C:\Users\#omitted user name#\Downloads\archive (1)\cloud',
      image_size=(SIZE, SIZE),
      batch_size=batch_size,
      label_mode=None

  )

  rescaling_layer = layers.Rescaling(1./255)


def change_inputs(images, labels=None):
  x = tensorflow.image.resize(rescaling_layer(images),[SIZE, SIZE], method=tensorflow.image.ResizeMethod.NEAREST_NEIGHBOR)
  return x, x


train_dataset = train_generator.map(change_inputs)
validation_dataset = validation_generator.map(change_inputs)
anomaly_dataset = anomaly_generator.map(change_inputs)

#some model building and compiling code goes here but I omitted it#

# Examine the recon. error between val data and anomaly images
validation_error = model.evaluate(validation_generator)
anomaly_error = model.evaluate(anomaly_generator)


# Print out the results
print(f"Recon. error for the validation data is {validation_error}")
print(f"Recon. error for the anomaly data is {anomaly_error}")

最后四行是由于 label_mode 导致的问题

Answer 1

您必须将 label_mode 设置为适当的类型，然后再转换